[SI-LIST] Re: Another question about the Initial Delays in IBIS

  • From: "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
  • To: <si-list@xxxxxxxxxxxxx>, <si-list@xxxxxxxxxxxxx>
  • Date: Thu, 4 Nov 2010 21:49:12 -0700

Liu,

We will have to get Lance to answer your email because I don't
know what model he used for the presentation, but I wonder 
whether he was observing how various simulators handle the
case when the stimulus is faster than the length of the V-t
curves.  This is also known as the over clocking situation,
which is not addressed by IBIS models.  I did some experiments
a long time ago with HSPICE and found interesting things:

http://www.eda.org/ibis/summits/jun03b/muranyi2.pdf

I hope this answers some of your questions...

Arpad
===============================================================

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On Behalf Of liuluping 41830
Sent: Thursday, November 04, 2010 9:11 PM
To: si-list@xxxxxxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Another question about the Initial Delays in IBIS

Hi all:
 
  I study Lauce Wang's "Initial Time Delay Issue in IBIS VT Curves" at
Designcon 2007:
 
http://www.cadence.com/rl/Resources/conference_papers/stp_wang_designcon
2007.pdf
  
  and have another questions about this topic:
  
  At page 6, when compare the simulation waveform of IBIS With full VT
waveforms and spice at 2ns bit width,
 
  The waveform of IBIS is narrower than the actual bit width, but as we
talked about before,it need to have enough 
  charge/discharge period for Die Capacitance when there is a long
"initial"delay in the IBIS VT curves.
  so why the IBIS simulation waveform is narrow than the actual bit
width but not by contraries ?
 
Best Regards, 
LIU Luping 
***************************************************************** 
   This e-mail and its attachments contain confidential information from
HUAWEI, which is intended only for the person 
or entity whose address is listed above. Any use of the information
contained herein in any way (including, but not 
limited to, total or partial disclosure, reproduction, or dissemination)
by persons other than the intended recipient(s) 
is prohibited. If you receive this e-mail in error, please notify the
sender by phone or email immediately and delete it! 
***********************************************
 
Msg: #1 in digest
From: "Pugazharasan Selvanathan - ERS, HCL Tech"
<pugazharasan.s@xxxxxxx>
Date: Sun, 24 Oct 2010 16:01:13 +0530
Subject: [SI-LIST] Re: Initial Delays in V-t Rise/Fall Curves in IBIS 

Hi Arpad,

Thank you so much for your reply, it did clear my doubts well. 

Pugazh

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On Behalf Of Muranyi, Arpad
Sent: Friday, October 22, 2010 7:30 PM
To: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Initial Delays in V-t Rise/Fall Curves in IBIS 

Pugazh,

The initial delays in V-t curves represent the delay across the buffer
from the time the stimulus switches to the time when the output changes.
This delay is not very useful because it all depends on how many of
the pre-driver circuitry is included in the transistor model from which
you are extracting the V-t curves.  We usually don't use this delay in
SI simulations, because flight times are usually calculated from a
waveform which is driven by a buffer loaded with the test load that is
used for the Tco specification of the buffer.

You can safely remove this delay time from the V-t tables, but be
careful
to remove the same absolute amount from all of the tables.  You cannot
remove different amounts because that will upset the on/off switching
relationships of the pullup and pulldown structures.

To answer why these delays are there in the IBIS models, I have to say
that it depends.  There are times when the model makers are forced to
remove these delays because otherwise the transition would not fit into
the bit time window and the IBIS model would become over-clocked.
When this is not a problem, it is simply a matter of choice to leave
these delays in the V-t tables.  It is easier not have to remove the
delay, so by default people would naturally leave it that way.

Removing this delay will not have an effect on Setup and Hold timings
at all (if those measurements are done correctly).

The delays do not "have to be" removed normally, only if the transition
time (edge rate) of the buffer is relatively slow with respect to the
bit time which would cause the model to be over clocked at its normal
operating frequency.

I hope this helps your understanding of the problem.

Arpad
========================================================================
===

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On Behalf Of Pugazharasan Selvanathan - ERS, HCL Tech
Sent: Friday, October 22, 2010 8:32 AM
To: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Initial Delays in V-t Rise/Fall Curves in IBIS 

Hi all,
Am doing timing analysis for DDR3 interface with Hyperlynx. I need to
validate the DDR3 controller and DRAM IBIS models before using them in
simulations. Both the models have some amount of initial time delays in
their V-t curves. I understand that this initial non-switching time
should be removed from the IBIS models before using them in timing
analysis for source synchronous interfaces like DDR3.

Have a few questions now...


1.       How much will be impact of this "removing initial delays" on
the setup and hold time margins?

2.       If the initial non-switching delay be removed, why it should be
specified in the IBIS model first? Excluding the relative delay between
different V-t tables.

3.       If the initial delays has to be removed, what should be the %
Voltage Threshold I should use?

Also, are there any standard timing specification for DDR3 Controller?
We are yet to receive that from the ASIC vendor, but wanted to go with
default timing spec. Right now we are in the process of delay matching
the addr./Clock/DQ/DQS group. So this information will of much help.

For strict timing analysis, should I have to include the package delays
also (Delay from die to pin) of DDR3 compliant DRAM in the timing margin
calculations?

Regards,
Pugazh



------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
 
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu
  

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
 
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu
  

Other related posts: